Response to Michele Baccarani

Correspondence

doubt that the observed differences could be due to chance. I understand that these experiments are difficult, expensive and time-consuming and I acknowledge that interesting data may be worth reporting even if they are still inconclusive. Incidentally, many reviewers would not share my opinion and would suggest rejection of papers unless they are conclusive. A reasonable compromise would be to acknowledge that the evidence is scant, if any, in the discussion, and to alter the title accordingly. However, I would like to profit by this study to make a more general case of the methods that are used to compare normal with leukemic, and of the underlying logic. In this study, as in many other studies, normal progenitors are identified, enumerated and tested based on assays that were developed specifically to identify and enumerate human normal progenitors, either CFU-GM or LTCIC. That makes sense. Leukemic progenitors were defined, identified, enumerated and tested based on the same assays. That makes less sense, because it is based on the assumption that ‘leukemic’ CFUGM and LTCIC are the biologic counter-party of normal CFU-

GM and LTCIC and that the organization of leukemic hemopoiesis is identical to the organization of normal hemopoiesis. In other words, are CFU-GM or LTCIC or whatever other progenitors, as they are defined using assays that were developed for normal cells, relevant to leukemia? In CML the number of ‘leukemic’ LTCIC is small and their ability to generate ‘leukemic’ CFU-GM is decreased, but Ph+ cells have no problems in outgrowing normal cells. In acute leukemia we are not able to identify ‘leukemic’ LTCIC, but leukemia grows as well. The progression of CML towards blastic phase can be decided and accomplished by cells that were never able to grow in a CFUGM or a LTCIC assay. Do you think that Leukemia might open a debate on these points? It could be interesting. M Baccarani

Cattedra di Ematologia Policlinico Universitario Piazza S Maria della Misericordia I-33100 Udine, Italy

Response to Michele Baccarani

TO THE EDITOR Dr Baccarani wonders whether the data presented in our paper ‘Hypersensitivity of bcr-abl-positive progenitors to hyperthermia in patients with chronic myeloid leukemia’ are firm enough to justify our statement that bcr-abl-positive progenitors are more sensitive to hyperthermic treatment than control cells. First, in contrast to what Dr Baccarani states, the differences between control and hyperthermia-treated samples as presented in Table 2 are significant not only for marrow CFUGM at 42°C, but also at the LTC-IC level both in marrow and in peripheral blood. It has to be admitted that unfortunately details of the table can be misinterpreted easily, since due to an error in the superscripts an ‘h’ in the LTC-IC column, which stood for significance, has been replaced by a meaningless ‘9’. Since, in addition, similar trends are seen in bone marrow at 43°C both at the CFU-GM and especially on the LTC-IC level, the only exception in the six evaluable comparisons concerns peripheral blood at 42°C at the CFU-GM level. The results of these clonogenic assays confirm the molecular analysis: the hyperthermia-induced shifts in relative numbers of bcr-abl-positive and -negative colonies as shown in Table 4 are present both at the CFU-GM and the LTC-IC level and where indicated are significant by the objective methods presented. It should be emphasized, however, that these shifts occur almost exclusively in those cases where the control samples already contain bcr-abl-negative colonies: in only one out of 11 cases where in control samples all colonies were bcr-abl positive, a bcr-abl-negative colony was observed. When considering the total number of cases in which indeed a purging effect was found, it should thus be kept in mind that of a total number of 29 available comparisons only in 18 cases did the control sample contain a mixture of bcr-abl-positive and -negative colonies. It is in these samples that a relative increase of bcr-abl-negative colonies Correspondence: GJ Ossenkoppele Received 10 March 1998; accepted 12 March 1998

was found in 15 out of 17 cases, be it that only in part of the individual cases was statistical significance reached. In our opinion this gives additional support for the conclusion that hyperthermia has a purging effect in CML with the restriction that such is seen almost exclusively in mixed colony samples. We certainly agree that a larger number of samples would strengthen our conclusions, but as pointed out in our paper, we aim at obtaining fresh chronic phase CML samples with mixtures of bcr-abl-positive and -negative progenitors and such samples are relatively rare, as was also acknowledged by Dr Baccarani. In a second comment, the point is raised whether CFU-GM and LTC-IC assays are relevant to investigate leukemia. Indeed it should be kept in mind that both the CFU-GM and the LTCIC assay are artificial in vitro assays with the latter having the additional complication that leukemic cells might propagate less efficiently in LTC-IC culture. Dr Baccarani suggests that the number of ‘leukemic’ LTC-IC as well as CFU-GM output in CML is too small to account for rapid overgrowth of normal cells by Ph+ cells. This is still a matter of debate. Our results indicate that the frequency of LTC-IC (number of LTCICs/CD34-positive cells) and the output (CFU-GM/LTC-IC) can be as high as in non-CML CD34 cells (Ref. 1, see for example also sample 1054 in Table 3, in which the colonies evaluated for frequency and output determination are 100% bcr-abl positive, as shown in Table 4). An additional point raised by the authors concerns blast crises CML and acute myeloid leukemia, a disease which is in some aspects very comparable to blast crises CML. These cells have probably undergone differentiation and one might argue that in vitro stem cell assays might be less appropriate for evaluation of in vitro treatments. On the other hand and in contrast to Dr Baccarani’s suggestion, it is our experience that LTC-ICs can be generated from acute myeloid leukemia, while in addition, using the NOD SCID mouse model, there is now support2 for the hypothesis that leukemia-initiating cells in most cases are present in a phenotypically defined CD34-positive subpopulation (CD38 negative) that also harbors the normal stem cells. Lastly, we

1169

Correspondence

1170

encourage any further debate on a controversial subject such as the use of in vitro stem/progenitor assays to extrapolate to clinical reality. SFT Thijsen GJ Schuurhuis JW van Oostveen GJ Ossenkoppele

Department of Hematology, Free University Hospital, BR238 De Boelelaan 1117 1081 HV Amsterdam The Netherlands

References 1 Thijsen SFT, Schuurhuis GJ, van Oostveen JW, Theijsmeijer AP, Niewint AWM, Ossenkoppele GJ. Mobilization of hematopoietic progenitors in patients with chronic myeloid leukemia. Bone Marrow Transplant 1997; 20: 835–842. 2 Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nature Med 1997; 3: 730–737.